On The Contrary | Episode 4: Is tech killing the planet?

Jim Fruchterman (Tech Matters) and Trisha Ray (Observer Research Foundation) discuss the hidden costs of adopting tech and AI to mitigate climate change, and what it means for people and the planet.

"So right now I would say that AI is being used mainly by the academic community, the scientific community, to do climate models, to understand that we do have a climate crisis, to keep measuring how things are, to forecast into the future—needs something like AI to help you do that. I would say that we’re weaker on the local modelling. So in other words, local weather monitoring and modelling isn’t that good across most of the world, it’s great in certain richer countries. So we know what’s going on at the world level. But often knowing what’s happening in your district, we may not have bothered to do that modelling. And I think that much of where I want to see AI being used is to help the local farmer, the local community leader, to actually understand better what’s going on, what might be going on, and help them make decisions within their own priorities and context. And I think AI is not there yet. Well, I think people are using technology more broadly. And a lot of ways to deal with climate, the biggest areas that you see is energy, people trying to innovate around energy, solar panels, you know, are not necessarily considered an advanced technology any more. But it’s a big part of this wind power. And I think that, for example, there was recently a billion-dollar commitment to buy the output of technology that took carbon out of the atmosphere, something that is not economic today". - Jim Fruchterman

"But the problem with how we talk about tech for climate is that we think it’s a clear-cut solution that will neatly fix all our problems. So that said these applications have a cost and nor can we or should we decouple our thinking about the solutions, as you mentioned, as well with how people and systems work. So let me explain that a little bit. One way in which we can see this problem is that it takes a lot of energy to train an AI model. And there’s a study that attempts to quantify the carbon emissions of machine learning. And it says that training a single natural language processing model emits as much carbon dioxide as a car in its entire lifetime. So that’s a lot. Now you’re seeing the idea of environmentally sustainable AI also kind of gradually coming into the mainstream. As Jim mentioned, there are some big tech companies that are already acting on this. UNESCO last year also adopted a recommendation on the ethics of AI that includes minimising the environmental impact of AI as one of the principles. But we have to remember that these developments are not happening in a vacuum, nor do they affect everybody in the same way". - Trisha Ray

The episode was produced by Maed in India.